GPT-5 is supposed to be nicer now

In a recent announcement, OpenAI revealed that it is rolling out an update for its latest AI model, GPT-5, aimed at making interactions feel more pleasant and inviting. The launch of GPT-5 faced some challenges, as CEO Sam Altman acknowledged, describing the rollout as "a little more bumpy than we’d hoped for." Many users expressed a preference for the previous version, GPT-4o, prompting OpenAI to respond with this update. The company has indicated that the changes are subtle yet effective, intending to create a more approachable experience for users. According to OpenAI, you can expect to see small, genuine affirmations in the model's responses, such as "Good question" or "Great start," which are intended to enhance user engagement without crossing into flattery. They assured that internal testing has shown no increase in sycophantic behavior compared to earlier versions of GPT-5. During a recent dinner event with journalists, OpenAI executives attempted to shift the conversation toward the future of the company beyond GPT-5. However, as reported by Max Zeff, the troubled launch remained a significant topic of discussion. OpenAI's Vice President Nick Turley commented on the model's straightforwardness, stating that while GPT-5 was previously "just very to the point," the update is set to infuse a warmer tone into its interactions.